Evergreen
How to Verify an Urgent Call When the Voice Sounds Real
Pause, leave the inbound channel, and independently contact the claimed person before sending money, codes, information, or access in response to an urgent call.
How to Verify an Urgent Call When the Voice Sounds Real
Pause. Do not send money, credentials, security codes, sensitive information, or account access. End or step away from the inbound conversation, find a contact route you trusted before the request arrived, and independently reach the person or another trusted person who can verify the situation.
Do not make the decision by judging whether the voice sounds real.
1. Stop the immediate action
An urgent request works by collapsing the time available to verify it. Do not transfer funds, buy gift cards, send cryptocurrency, reveal a one-time code, click a supplied link, install software, or move to a new platform while the request remains unverified.
The FTC's family-emergency guide identifies urgency, secrecy, emotional pressure, and difficult-to-reverse payments as recurring warning signs.
2. Leave the inbound channel
Do not use the callback number, link, account, directory, or contact information supplied by the same message.
End the call or stop responding. Open your own contacts, a previously confirmed conversation, an official directory you locate independently, or another route established before the request.
Leaving the channel matters because the person making the request may control the caller ID, account, link, or proposed platform.
3. Contact the person through a known route
The FTC's AI voice-cloning warning says not to trust the voice. It recommends calling the supposed sender through a number already known to be theirs.
If the person does not answer, contact another family member, friend, colleague, security owner, or trusted person who can verify the situation through their own route.
An unanswered call is not confirmation of the emergency. Continue to withhold the requested action until the situation is independently verified.
4. Verify identity and event separately
Reaching the person can verify identity while leaving the claimed event unresolved. A compromised account or coerced person can still deliver a harmful request.
Ask whether the event occurred, whether the person made the request, what action is actually needed, and whether the payment or access method is legitimate. For an organization, use its established financial, security, legal, or incident process.
flowchart TD
A["Urgent familiar-sounding request"] --> B["Pause and withhold action"]
B --> C["Leave the inbound channel"]
C --> D["Use a previously trusted route"]
D --> E{"Reach the claimed person?"}
E -->|Yes| F["Verify identity, event, and requested action separately"]
E -->|No| G["Contact a second trusted person or official channel"]
F --> H["Act only through the verified process"]
G --> H
5. Treat familiar signals as insufficient
| Signal | Why it can persuade | Why it does not settle identity |
|---|---|---|
| Familiar voice | Resembles a person you know | It can be generated, replayed, edited, or performed |
| Caller ID | Displays a saved name or number | Numbers can be spoofed and accounts can be compromised |
| Personal details | Includes travel, family, work, or health context | Public posts, breaches, prior messages, and compromised accounts can reveal it |
| Video | Shows a recognizable face | Video can be generated, replayed, manipulated, or out of context |
| Urgency | Makes delay feel harmful | Pressure prevents independent verification |
| Secrecy | Frames disclosure as dangerous or embarrassing | Isolation blocks the people most likely to detect the deception |
| Audible flaw | Sounds delayed, flat, or unnatural | Real calls fail, and synthetic speech can sound natural |
The FBI's May 2025 alert documents a specific campaign using texts and AI-generated voice messages. The FBI recommends independently identifying contact information and verifying through a previously confirmed platform or trusted source.
That campaign is one example. Its independent-contact principle applies more broadly than its target group.
6. Use a family phrase only as an extra layer
A private phrase can add friction if everyone understands when to use it and it has not been published, texted into an exposed account, or included in a public example.
It is not sufficient for a transfer, account change, health decision, travel emergency, or other high-stakes action. The phrase can be overheard, disclosed, recorded, coerced, guessed, or reused.
If the phrase may have been exposed, replace it. Never publish the real phrase in a family plan, article, screenshot, or shared template.
7. Preserve and report without improvising
Keep the message, number, account, URL, payment instructions, timestamp, and relevant screenshots without continuing the suspicious interaction.
Use current official guidance for reporting and recovery. The FTC links its ReportFraud process from its consumer warning. If money or account access has already moved, contact the relevant financial institution, platform, employer, or official response channel immediately through independently located contact information.
Facts and jurisdiction can change the right recovery or reporting action. Do not rely on a public article as emergency, legal, financial, or law-enforcement advice.
Prepare before the call arrives
Agree on the phone numbers and platforms that count as known routes. Choose a second trusted contact. Decide what no one in the family or team will request under pressure. Keep financial and account recovery channels current.
Practice the route once while nothing is wrong. The exercise should confirm that everyone can find the trusted number, reach the backup person, and pause without feeling that delay itself violates the family or team's expectations.
The strongest plan does not require anyone to detect a fake voice. It gives them permission to pause and a separate path to the truth.
This page was developed with AI assistance and reviewed against current FTC and FBI guidance and the preserved E063 episode linked above. Dalton Anderson is responsible for the final editorial judgment. Recheck official reporting and recovery guidance for the actual facts before acting.
Sources
Follow the evidence.
- copyright.gov: Copyright and Artificial Intelligence Part 1 Digital Replicas Reportcopyright.gov
- elevenlabs.io: voice cloningelevenlabs.io
- consumer.ftc.gov: scammers use ai enhance their family emergency schemesconsumer.ftc.gov
- reportfraud.ftc.govreportfraud.ftc.gov
- open.spotify.com: 4gr8yx2FQB0taJ0dhF0DLbopen.spotify.com
- fbi.gov: senior us officials impersonated in malicious messaging campaignfbi.gov
- spec.c2pa.org: ContentCredentialsspec.c2pa.org
- docs.fcc.gov: DOC 400393A1docs.fcc.gov
- consumer.ftc.gov: scammers use fake emergencies steal your moneyconsumer.ftc.gov
- nist.gov: reducing risks posed synthetic content overview technical approaches digital contentnist.gov
- spec.c2pa.org: charterspec.c2pa.org
- youtu.be: AW eZuxKf Myoutu.be
- docs.fcc.gov: FCC 24 17A1docs.fcc.gov
- copyright.gov: aicopyright.gov
- daltonanderson.ghost.io: the imperfect echo ai voice cloning digital trustdaltonanderson.ghost.io